def main_train(pos_sequences=None,
neg_sequences=None,
prefix=None,
model_file=None,
weights_file=None):
# encode fastas
print("loading sequence data...")
X_pos = encode_fasta_sequences(pos_sequences)
y_pos = np.array([[True]]*len(X_pos))
X_neg = encode_fasta_sequences(neg_sequences)
y_neg = np.array([[False]]*len(X_neg))
X = np.concatenate((X_pos, X_neg))
y = np.concatenate((y_pos, y_neg))
X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.2)
if model_file is not None and weights_file is not None: # load model
print("loading model...")
model = SequenceDNN.load(model_file, weights_file)
else: # initialize model
print("initializing model...")
model = SequenceDNN(seq_length=X_train.shape[-1])
# train
print("starting model training...")
model.train(X_train, y_train, validation_data=(X_valid, y_valid))
valid_result = model.test(X_valid, y_valid)
print("final validation metrics:")
print(valid_result)
# save
print("saving model files..")
model.save("%s.model.json" % (prefix), "%s.weights.hd5" % (prefix))
print("Done!")
def main_train(pos_sequences=None,
neg_sequences=None,
prefix=None,
arch_file=None,
weights_file=None,
**kwargs):
kwargs = {key: value for key, value in kwargs.items() if value is not None}
# encode fastas
print("loading sequence data...")
X_pos = encode_fasta_sequences(pos_sequences)
y_pos = np.array([[True]] * len(X_pos))
X_neg = encode_fasta_sequences(neg_sequences)
y_neg = np.array([[False]] * len(X_neg))
X = np.concatenate((X_pos, X_neg))
y = np.concatenate((y_pos, y_neg))
X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.2)
if arch_file is not None: # load model
print("loading model...")
model = SequenceDNN.load(arch_file, weights_file)
else: # initialize model
print("initializing model...")
model = SequenceDNN(seq_length=X_train.shape[-1], **kwargs)
# train
print("starting model training...")
model.train(X_train, y_train, validation_data=(X_valid, y_valid))
valid_result = model.test(X_valid, y_valid)
print("final validation metrics:")
print(valid_result)
# save
print("saving model files..")
model.save(prefix)
print("Done!")
def main_train(pos_sequences=None,
neg_sequences=None,
prefix=None,
arch_file=None,
weights_file=None,
**kwargs):
kwargs = {key: value for key, value in kwargs.items() if value is not None}
# encode fastas
print("loading sequence data...")
X_pos = encode_fasta_sequences(pos_sequences)
y_pos = np.array([[True]]*len(X_pos))
X_neg = encode_fasta_sequences(neg_sequences)
y_neg = np.array([[False]]*len(X_neg))
X = np.concatenate((X_pos, X_neg))
y = np.concatenate((y_pos, y_neg))
X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.2)
if arch_file is not None: # load model
print("loading model...")
model = SequenceDNN.load(arch_file, weights_file)
else: # initialize model
print("initializing model...")
model = SequenceDNN(seq_length=X_train.shape[-1], **kwargs)
# train
print("starting model training...")
model.train(X_train, y_train, validation_data=(X_valid, y_valid))
valid_result = model.test(X_valid, y_valid)
print("final validation metrics:")
print(valid_result)
# save
print("saving model files..")
model.save(prefix)
print("Done!")
def main_train(pos_sequences=None,
neg_sequences=None,
pos_validation_sequences=None,
neg_validation_sequences=None,
prefix=None,
arch_file=None,
weights_file=None,
**kwargs):
kwargs = {key: value for key, value in kwargs.items() if value is not None}
# encode fastas
print("loading sequence data...")
X_pos = encode_fasta_sequences(pos_sequences)
y_pos = np.array([[True]] * len(X_pos))
X_neg = encode_fasta_sequences(neg_sequences)
y_neg = np.array([[False]] * len(X_neg))
X = np.concatenate((X_pos, X_neg))
y = np.concatenate((y_pos, y_neg))
#if a validation set is provided by the user, encode that as well
if (pos_validation_sequences != None or neg_validation_sequences != None):
#both positive and negative validation sequences must be provided.
assert neg_validation_sequences != None
assert pos_validation_sequences != None
X_valid_pos = encode_fasta_sequences(pos_validation_sequences)
X_valid_neg = encode_fasta_sequences(neg_validation_sequences)
y_valid_pos = np.array([[True]]) * len(X_valid_pos)
y_valid_neg = np.array([[False]]) * len(X_valid_neg)
X_valid = np.concatenate((X_valid_pos, X_valid_neg))
y_valid = np.concatenate((y_valid_pos, y_valid_neg))
else:
X_train, X_valid, y_train, y_valid = train_test_split(X,
y,
test_size=0.2)
if arch_file is not None: # load model
print("loading model...")
model = SequenceDNN.load(model_hdf5_file, arch_file, weights_file)
else: # initialize model
print("initializing model...")
model = SequenceDNN(seq_length=X_train.shape[-1], **kwargs)
# train
print("starting model training...")
model.train(X_train, y_train, validation_data=(X_valid, y_valid))
valid_result = model.test(X_valid, y_valid)
print("final validation metrics:")
print(valid_result)
# save
print("saving model files..")
model.save(prefix)
print("Done!")
def main_test(pos_sequences=None,
neg_sequences=None,
arch_file=None,
weights_file=None):
# encode fastas
print("loading sequence data...")
X_test_pos = encode_fasta_sequences(pos_sequences)
y_test_pos = np.array([[True]] * len(X_test_pos))
X_test_neg = encode_fasta_sequences(neg_sequences)
y_test_neg = np.array([[False]] * len(X_test_neg))
X_test = np.concatenate((X_test_pos, X_test_neg))
y_test = np.concatenate((y_test_pos, y_test_neg))
# load model
print("loading model...")
model = SequenceDNN.load(arch_file, weights_file)
# test
print("testing model...")
test_result = model.test(X_test, y_test)
print(test_result)
def main_test(pos_sequences=None,
neg_sequences=None,
model_file=None,
weights_file=None):
# encode fastas
print("loading sequence data...")
X_test_pos = encode_fasta_sequences(pos_sequences)
y_test_pos = np.array([[True]]*len(X_test_pos))
X_test_neg = encode_fasta_sequences(neg_sequences)
y_test_neg = np.array([[False]]*len(X_test_neg))
X_test = np.concatenate((X_test_pos, X_test_neg))
y_test = np.concatenate((y_test_pos, y_test_neg))
# load model
print("loading model...")
model = SequenceDNN.load(model_file, weights_file)
# test
print("testing model...")
test_result = model.test(X_test, y_test)
print(test_result)
def main(args):
'''
args - parsed arguments that include pos_sequences, neg_sequences,
arch_file, and weights_file
'''
# encode fasta
print('Loading sequence data...')
pos_seq = encode_fasta_sequences(args.pos_sequences)
print('{} positive test sequences'.format(len(pos_seq)))
neg_seq = encode_fasta_sequences(args.neg_sequences)
print('{} negative test sequences\n'.format(len(neg_seq)))
# load model
prefix = args.arch_file.replace('.arch.json', '')
print('Loading {} model...'.format(prefix))
model = SequenceDNN.load(args.arch_file, args.weights_file)
# predict binding probability on test sequences
print('Getting predictions...')
pos_predictions = model.predict(pos_seq)
for index, pred in enumerate(pos_predictions):
print('positive_test_{}\tP(bound)={}'.format(index, pred[0]))
print('')
neg_predictions = model.predict(neg_seq)
for index, pred in enumerate(neg_predictions):
print('negative_test_{}\tP(bound)={}'.format(index, pred[0]))
print('')
# visualize trained model and motifs
print('Plotting deeplift scores on positive sequences...')
model.plot_deeplift(pos_seq, '{}_deeplift_positive'.format(prefix))
print('Plotting true motifs...')
motif_names = ['IRF_known1', 'NFKB_known1']
for index, motif in enumerate(motif_names):
fig = plot_motif(motif, figsize=(10, 4), ylab=motif)
fig.savefig('motif{}.png'.format(index + 1), bbox_inches='tight')
print('Plotting architecture...')
model.plot_architecture('{}_architecture.png'.format(prefix))
print('Plotting convolutional filters...')
plot_sequence_filters(model, prefix)
def main_predict(sequences=None,
arch_file=None,
weights_file=None,
output_file=None):
# encode fasta
print("loading sequence data...")
X = encode_fasta_sequences(sequences)
# load model
print("loading model...")
model = SequenceDNN.load(arch_file, weights_file)
# predict
print("getting predictions...")
predictions = model.predict(X)
# save predictions
print("saving predictions to output file...")
np.savetxt(output_file, predictions)
print("Done!")
def main_predict(sequences=None,
model_file=None,
weights_file=None,
output_file=None):
# encode fasta
print("loading sequence data...")
X = encode_fasta_sequences(sequences)
# load model
print("loading model...")
model = SequenceDNN.load(model_file, weights_file)
# predict
print("getting predictions...")
predictions = model.predict(X)
# save predictions
print("saving predictions to output file...")
np.savetxt(output_file, predictions)
print("Done!")
def main_interpret(sequences=None,
arch_file=None,
weights_file=None,
pos_threshold=None,
peak_width=10,
prefix=None):
# encode fasta
print("loading sequence data...")
X = encode_fasta_sequences(sequences)
# load model
print("loading model...")
model = SequenceDNN.load(arch_file, weights_file)
# predict
print("getting predictions...")
predictions = model.predict(X)
# deeplift
print("getting deeplift scores...")
deeplift_scores = model.deeplift(X)
# get important sequences and write to file
print("extracting important sequences and writing to file...")
for task_index, task_scores in enumerate(deeplift_scores):
peak_positions = []
peak_sequences = []
for sequence_index, sequence_scores in enumerate(task_scores):
if predictions[sequence_index, task_index] > pos_threshold:
#print(sequence_scores.shape)
basewise_sequence_scores = sequence_scores.max(axis=(0,1))
peak_position = basewise_sequence_scores.argmax()
peak_positions.append(peak_position)
peak_sequences.append(X[sequence_index : sequence_index + 1,
:,
:,
peak_position - peak_width :
peak_position + peak_width])
else:
peak_positions.append(-1)
peak_sequences.append(np.zeros((1, 1, 4, 2 * peak_width)))
peak_sequences = np.concatenate(peak_sequences)
peak_sequence_strings = get_sequence_strings(peak_sequences)
# write important sequences to file
ofname = "%s.task_%i.important_sequences.txt" % (prefix, task_index)
with open(ofname, "w") as wf:
for i, peak_position in enumerate(peak_positions):
wf.write("> sequence_%i\n" % (i))
wf.write("%i: %s\n" %(peak_position, peak_sequence_strings[i]))
print("Done!")
def main_interpret(sequences=None,
arch_file=None,
weights_file=None,
pos_threshold=None,
peak_width=10,
prefix=None):
# encode fasta
print("loading sequence data...")
X = encode_fasta_sequences(sequences)
# load model
print("loading model...")
model = SequenceDNN.load(arch_file, weights_file)
# predict
print("getting predictions...")
predictions = model.predict(X)
# deeplift
print("getting deeplift scores...")
deeplift_scores = model.deeplift(X)
# get important sequences and write to file
print("extracting important sequences and writing to file...")
for task_index, task_scores in enumerate(deeplift_scores):
peak_positions = []
peak_sequences = []
for sequence_index, sequence_scores in enumerate(task_scores):
if predictions[sequence_index, task_index] > pos_threshold:
#print(sequence_scores.shape)
basewise_sequence_scores = sequence_scores.max(axis=(0, 1))
peak_position = basewise_sequence_scores.argmax()
peak_positions.append(peak_position)
peak_sequences.append(
X[sequence_index:sequence_index + 1, :, :,
peak_position - peak_width:peak_position + peak_width])
else:
peak_positions.append(-1)
peak_sequences.append(np.zeros((1, 1, 4, 2 * peak_width)))
peak_sequences = np.concatenate(peak_sequences)
peak_sequence_strings = get_sequence_strings(peak_sequences)
# write important sequences to file
ofname = "%s.task_%i.important_sequences.txt" % (prefix, task_index)
with open(ofname, "w") as wf:
for i, peak_position in enumerate(peak_positions):
wf.write("> sequence_%i\n" % (i))
wf.write("%i: %s\n" %
(peak_position, peak_sequence_strings[i]))
print("Done!")
